Remote sensing indices and their applications

Dr. Poonam Tripathi

Training Analyst, Capacity Building
ICIMOD

Training of Trainers on the applications of remote sensing and GIS for Afghanistan
9-13 March, 2020
Electromagnetic spectrum (EMR)
From very short Gamma rays to very long radio waves

Passive - optical Active - microwave

Gamma X-rays Ultraviolet Visible Infrared Microwave Radio-wave

10-14 10-12 10-8 10-6 10-4 10-2 1 l in meter

400 500 600 700 l in nanometer

©2017 Gauri

Blue (400 – 500), Green (500 – 600) and Red (600 – 700nm) bands
Features in Sentinel-2A satellite bands
Spectral
/wavelength Spatial
Band range (nm) Objective range (m)
Kabul region of Afghanistan
B1 Coastal aerosol 433-453 Aerosol correction 60
Aerosol correction,
B2 Blue 458-523 land measurement 10
B3 Green 543-578 Land measurement 10
B4 Red 650-680 Land measurement 10
B5 Red edge1 (RE1) 698-713 Land measurement 20
Google Earth B2 (Blue) B3 (Green) B6 Red edge2 (RE2) 733-748 Land measurement 20
B7 Red edge3 (RE3) 773-793 Land measurement 20
Water vapour correction,
B8 Near infra red 785-900 Land measurement 10
Water vapour correction,
B8a Near infrared narrow 855-875 Land measurement 20
B9 Water vapour 935-955 Water vapour correction 60
B4 (Red) B8 (NIR) B9 (SWIR) B10 Shortwave infrared 1360-1390 Cirrus detection 60
B11 Shortwave infrared 1 1565-1655 Land measurement 20
Aerosol correction,
B12 Shortwave infrared 2 2100-2280 land measurement 20
Features in Landsat satellite bands
Spectral Spatial
Band range (nm) Objective range (m)
B1 New deep blue 433-453 Aerosol/coastal zone 30
B2 Blue 450-515 Pigments/coastal/scatter 30
B3 Green 525-600 Pigments/coastal 30
B4 Red 630-680 Pigments/coastal 30
B5 Near infra red 845-885 Foliage/coastal 30
B6 Shortwave infrared 2 1560-1660 Foliage 30
B7 Shortwave infrared 3 2100-2300 Mineral/litter/no scatter 30
B8 Panchromatic 500-680 Image sharpening 15
B9 Shortwave infrared 1360-1390 Cirrus cloud detection 30

1. https://www.sentinel-hub.com/develop/documentation/eo_products/Sentinel2EOproducts
2. https://modis.gsfc.nasa.gov/about/specifications.php
3. https://gisgeography.com/landsat-8-bands-combinations/
Differences between Landsat and Sentinel data
Landsat-8-OLI Sentinel-2A Landsat-8-OLI

Bands 9 13
Spectral range (μm) 0.435-1.384 0.44-2.22
Spatial resolution (m) 30 10,20,60
Temporal resolution 16 days 10 days
Operational Land Multi-Spectral
Sentinel-2A
Sensor Imager (OLI) Instrument (MSI)
Type Multi-spectral Multi-spectral
Satellite Landsat-8 Sentinel-2A
U.S. Geological European Space
Operator Survey (USGS) Agency (ESA)
Why band ratio?
 Undesirable effects on recorded radiances (e.g. variable
illumination) caused by variation in topography
 Differences in brightness values from identical surface
material or vice versa are caused by topographic slope and
aspect, shadows or seasonal changes
 These hamper the ability of interpreter to correctly identify
surface material in image
 Ratio transformation can be used to reduce the effects of such
environmental conditions
Why band ratio?
Spectral response and
reflectance is similar from
two different objects in
RED band of Sentinel-2A
Vegetation 732
Google Earth
Google Earth Red band
Water
725

Red band
Why band ratio?
Google Earth NIR band

Vegetation
3332

Fallow land 3351

Spectral signature of two different objects are similar in NEAR-INFRA

RED band of Sentinel
Why band ratio?
Google Earth NIR band

Snow
3585
Cloud 3557

Spectral signature of two different objects are similar in NEAR-INFRA RED band
Why band ratio?

Specific target has an

individual and characteristic
manner of interacting with
incident radiation
Interaction are described by
the spectral response of
the target in a particular
wavelength of EMR
Spectral curves for various natural features
Differences between histogram
Red

Histogram of Red and near-infrared

reflectance representing more pixel
frequency at higher reflectance in NIR of
Sentinel-2A data

NIR

GEE B4 (Red) B6 (NIR)

Spectral indices

1. Spectral indices help in modelling, predicting, or infer surface processes

2. Developed to assess and monitor several land change processes
3. Computed from multiband images by adding and subtracting bands
thereby making various band ratio
4. Emphasizes a specific phenomenon that is present, while mitigating
other factors
• Vegetation health and status
• Burned area
• Fire severity etc.
Development of spectral indices
1. Initially intrinsic indices were developed from simple band ratios,
which highlighted the spectral properties of vegetation at different
stages of growth and senescence.
2. To compensate for background effects such as that caused in areas in
which the soil response dominates over the vegetation.
3. To compensate for the effects of atmospheric distortion.
4. Development of new spectral indices to applications other than
vegetation health. These include indices for burned area assessment
and fire severity etc.
The criterion of a spectral index
•Maximize the sensitivity of certain surface feature (e.g. plant biophysical
properties). Ideally, such responses should change linearly to allow both ease
of scaling and use over a wide range of surface conditions.
•Normalize or reduce effects due to sun angle, viewing angle, the
atmosphere, topography, instrument noise, etc., to allow consistent spatial and
temporal comparisons
•Be linked to specific and measurable surface processes (e.g. biophysical
parameter such as leaf area index (LAI), biomass, absorbed photosynthetically
active radiation (APAR, etc.)) – i.e. be related to a measurable parameter or
process
Source: Jenson (RSE Book, 2000)
Use of ratio to reduce topographic effects
Example 1
Use of indices to reduce topographic effects
Example 2 from Kabul Normalized difference
False color composite (FCC) vegetation indices (NDVI)

R: 703 Sunlit: 0.36

NIR: 1502 Shaded:0.359
R:480
NIR:1018

R and NIR represents the reflectance in red and near infra-red band of Sentinel
for SUNLIT and SHADED region and their respective ratio in NDVI image
Vegetation indices
 Enhances green vegetation so that plants appear distinct
from other image features
 Reflectance of light spectra from plants changes with plant
type, water content within tissues, and other intrinsic
factors
 Vegetation reflectance is determined by chemical and
morphological characteristics of the surface of organs or
leaves e.g. leaf structure, leaf pigments etc.
Vegetation indices

Leaf structure and its

responses in different
EMR region
Vegetation indices
 Simple Difference Vegetation Index
DVI = NIR-R (Richardson et al., 1977)
DVI
– Distinguishes between soil and vegetation
– Does Not deal with the atmospheric effects

 Ratio-based Vegetation Index

RVI = R/NIR (Jordan 1969 )
– Reduces the effects of atmosphere and RVI
topography
– Low for soil, ice, water, etc.
Vegetation indices
 Normalized Difference Vegetation
Index
 Standard method for comparing the
vegetation greenness from satellite

 NDVI = (NIR)-(RED)/ (NIR)+ (RED)

 Explains density of vegetation

 The NDVI values tentatively ranges

between -1 to +1, the values close to +1
denotes the good health of vegetation

NDVI is sensitive to the effects of soil and atmosphere and saturates at high density of vegetation
Vegetation indices
NDVI

 Negative values correspond to areas with water surfaces, manmade structures,

rocks, clouds, snow;
 Plants will always have positive values between 0.2 and 1.
Vegetation indices
 Soil Adjusted Vegetation Index
 The concept of distinction of vegetation from the soil background was proposed by
Richardson and Wiegand., 1977
 Soil background conditions exert considerable influence on partial canopy spectra
and the calculated vegetation indices
 SAVI = ((NIR – Red) / (NIR + Red + L)) x (1 + L) Huete, 1988
 Minimizes soil brightness influence
 L is a variable ranges within -1 to 1, depending on the amount of green vegetation
present in the area
 To run the remote sensing analysis of areas with high green vegetation, L is set to be
zero (in which case SAVI index data will be equal to NDVI); whereas low green
vegetation regions require L=1
Vegetation indices
Atmospherically Resistant Vegetation Index (ARVI; Kaufman and
Tanré, 1992)
 Relatively prone to atmospheric factors (such as aerosol)
 Atmosphere affects significantly Red region compared to the NIR
 Corrected for atmospheric scattering effects in the red reflectance
spectrum by using the measurements in blue wavelengths.
ARVI = (NIR – (2 * Red) + Blue) / (NIR + (2 * Red) + Blue)
 Eliminates the effect of atmospheric aerosols
 Monitoring tool for tropical mountainous regions often polluted by soot
coming from slash-and-burn agriculture
Vegetation indices
 Land Surface Water Index (LSWI)
NIR-SWIR/NIR+SWIR
LSWI is sensitive to changes in vegetation
canopy water content and indicates the
water stress

 Vegetation Condition Index (VCI)

LSWI
(NDVI-NDVImin/NDVImax- NDVImin)*100
Lower and higher values indicate bad and
good vegetation state conditions
Water indices
 Used to highlight the water bodies while suppressing the other land cover
 Water absorbs more energy (low reflectance) in NIR and SWIR
wavelengths
 Have the greatest reflectance in the blue portion of the visible spectrum
 Clear water has high absorption and virtually no reflectance in near
infrared wavelengths range and beyond

Factors affecting water

 Algae: Water with higher algal density reflect more in green bands
 Turbidity: Turbid water has a higher reflectance in the visible region than clear
water. This is also true for waters containing high chlorophyll concentrations
Water indices

Adapted from Kristi

Uudeberg, 2019

Reflectance response of water with different levels of turbidity

Snow indices
Blue Green Red
 Ice and snow generally have high
reflectance across all visible
wavelengths, thus bright white
appearance
NIR SWIR FCC
 The low reflection of ice and
snow in the SWIR is related to
their microscopic liquid water
content
Snow indices

Source: USGS

Spectral reflectance curves of bare glacier ice, coarse-grained snow, and fine-grained
snow. Spectral bands of selected sensor on Earth-orbiting satellites are shown in gray.
The numbers in the gray boxes refer to the associated band numbers of each sensor
Water and snow indices
Indices Formula

Normalized Difference Water Index (NDWI) GREEN-NIR/GREEN+NIR

Modified Normalized Difference Water Index (MNDWI) GREEN-SWIR/GREEN+SWIR
Normalized Difference Pond Index (NDPI) MIR-GREEN/MIR+GREEN
Water Ration Index (WRI) GREEN+RED/NIR+SWIR

Normalized Difference Turbidity Index (NDTI) RED-GREEN/RED+GREEN

4*(GREEN-SWIR2-
Automated Water Extraction Index (AWEI) 0.25*NIR+2.75*SWIR1

Normalized Difference Snow Index (NDSI) GREEN-SWIR/GREEN+SWIR

Normalized Difference Snow and Ice Index (NDSII-1) RED-SWIR/RED+SWIR
GREEN (NIR-SWIR)/
Snow Water Index (SWI) (GREEN+NIR)(NIR+SWIR)
Applications
Application of vegetation indices
 Vegetation mapping and monitoring
 Biodiversity assessment
 Estimation of biophysical parameters (LAI, fPAR)
 Phenological assessment
 Vegetation health/stress
 Forest degradation
 Biomass mapping and modelling
 Productivity and carbon assessment
 Crop condition monitoring and predicting crop yield
Application of vegetation indices

Source: Hatfield and Prueger (2010)

Application of vegetation indices

Leaf area
index (LAI)
estimation

Plantation R2 a B c F-value
a) Mixed 0.59 0.06 2.47 2.13 49.8

b) Eucalyptus 0.47 0.07 2.73 1.52 21.6

c) Poplar 0.46 0.05 3.13 2.26 20.4

d) All plantation 0.62 0.05 2.73 2.01 63.3

Source: Tripathi et al., 2014

Application of vegetation indices
January

May

September

Monthly pattern (January- December) of leaf area index

(LAI) derived from NDVI (exponential relationship)
Application of vegetation indices
Biomass estimation

Correlation between above ground biomass and

vegetation indices for Quercus rotundifolia

Source: Macedo et al., 2018, southern Portugal

Application of vegetation indices
Drought Assessment

Source: Thenkabail and Gamage, 2004

Application of vegetation indices

Southwest Asia
NDVI deviation

 A monthly NDVI time series

for a drought year (1987)
and a wet year (1993)
compared to the NDVI long-
term mean

Source: Thenkabail and Gamage, 2004

Application of vegetation indices

Monthly median of drought condition in Herat province

during vegetation seasons of 2003-2014 based on VCI
Source: Mohammad Ehsan Razipoor, 2019
http://www.fao.org/giews/earthobservation/asis/index_2.jsp?lang=en
Application of water and snow indices
 Water mapping and monitoring
 Change detection
 Water quality assessment
 Flood monitoring and damage assessment
 Algae assessment
 Snow and ice mapping and monitoring
Application of water indices
(b) (c)
Water mapping
(a)

(d) (e)

Source: Sharma et al, 2015

(a) Reference water cover map (b) Automated water extraction index (AWEI), (c) Modified normalized difference water
index (MNDWI) (d) Normalized difference water index (NDWI), (e) Superfine water index (SWI)
Application of water indices

Water mapping and

change dynamics

Simple water index (SWI)

Source: Acharya et al., 2019

Application of water indices
Water inundation mapping
 Minimum and maximum inundation
areas and their distributions during
each climatological month between
2000 and 2010 for Poyang Lake,
China

Source: Feng et al. (2012)

Application of water indices
Water quality assessment

Source: Ahmed M. El-Zeiny, 2018

Application of water indices
Flood hotspot analysis
International Water Management Institute project

Source: Matheswaran et al, 2018

Application of snow indices

Source: Bajracharya et al., 2015

Thank You
E-mail: Poonam. [email protected]